Hi all:

   This is getting really dangerous... I managed to fix the segfaults on
Koppi's bullet physics playground, and now the fun has no end. I even
managed to add bare mesh support, as the lib3ds code was there and
almost finished, and this opens a whole universe of "now possible"
scenes. I don't know how I'm going to get any serious work done from now
on...

   For the moment there are no compound nor constrains, just rigid meshes
linked together because of the geometry and careful placement: the
simulation does it's job, and the links work, even if sometimes break
due to accuracy problems (still haven't decided if it's a bug or a feature).

   Let's see if I can stop playing, so I can fork Koppi's code to create
a more POV-oriented version: I'm envisioning some simple features that
will make the life of any POV user much easier...

--
Jaime

Post a reply to this message

Attachments:
Download 'chain-fall-04.jpg' (123 KB) Download 'chains-curtain-01.jpg' (156 KB) Download 'pearls-collar-05.jpg' (276 KB)

Preview of image 'chain-fall-04.jpg'

Preview of image 'chains-curtain-01.jpg'

Preview of image 'pearls-collar-05.jpg'

On 15-10-2012 14:49, Jaime Vives Piqueres wrote:
>    Let's see if I can stop playing, so I can fork Koppi's code to create
> a more POV-oriented version: I'm envisioning some simple features that
> will make the life of any POV user much easier...

By the tits of the Satrap! (Sorry, I should not use the Satrap's name in 
blasphemy. Very unhealthy) I already can see this in the harbour of 
Gancaloon!

Thomas

Post a reply to this message

>Jaime Vives Piqueres  on date 15/10/2012 14.49 wrote:
> Hi all:
>
>    This is getting really dangerous... I managed to fix the segfaults on
> Koppi's bullet physics playground, and now the fun has no end. I even
> managed to add bare mesh support, as the lib3ds code was there and
> almost finished, and this opens a whole universe of "now possible"
> scenes. I don't know how I'm going to get any serious work done from now
> on...
>
>    For the moment there are no compound nor constrains, just rigid meshes
> linked together because of the geometry and careful placement: the
> simulation does it's job, and the links work, even if sometimes break
> due to accuracy problems (still haven't decided if it's a bug or a
> feature).
>
>    Let's see if I can stop playing, so I can fork Koppi's code to create
> a more POV-oriented version: I'm envisioning some simple features that
> will make the life of any POV user much easier...
>
> --
> Jaime
Please, stop post photos here...
;-)
Paolo

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> Hi all:
>
>    This is getting really dangerous... I managed to fix the segfaults on
> Koppi's bullet physics playground, and now the fun has no end. I even
> managed to add bare mesh support, as the lib3ds code was there and
> almost finished, and this opens a whole universe of "now possible"
> scenes. I don't know how I'm going to get any serious work done from now
> on...
>
>    For the moment there are no compound nor constrains, just rigid meshes
> linked together because of the geometry and careful placement: the
> simulation does it's job, and the links work, even if sometimes break
> due to accuracy problems (still haven't decided if it's a bug or a feature).
>
>    Let's see if I can stop playing, so I can fork Koppi's code to create
> a more POV-oriented version: I'm envisioning some simple features that
> will make the life of any POV user much easier...
>
> --
> Jaime

Wow, these are really impressive. I want to start playing around with Koppi's
bullet physics playground. Anyone have tips on getting it working on OS X?

Henry

Post a reply to this message

On 15/10/12 18:43, HenryW wrote:
> I want to start playing around with Koppi's bullet physics
> playground. Anyone have tips on getting it working on OS X?

   I don't recommend it, as you can reach insane levels of fun, and mean
REALLY insane.

   Anyhow, if you want to try it, you may need to put some hard work to
get it compiled. The author has not wrote anything into the wiki howto,
but seems he plans to... the problems is that he is unreachable until
mid-december. I really don't know for sure, but I think all the
libraries used in the project must be available for OS X, and gcc too,
so it should be relatively easy if you already have experience compiling
from sources on OS X.

   In any case, if you manage to compile it, then contact me for the
fixes I did to the current code on github, else you will not be able to
do much with it.

  Regards,

--
Jaime

Post a reply to this message

On 15/10/12 19:36, Jaime Vives Piqueres wrote:
> I don't recommend it, as you can reach insane levels of fun, and
> mean REALLY insane.

   See the attached screen-shot from the GL preview: do you see what I
mean? ;)

--
Jaime

Post a reply to this message

Attachments:
Download 'pantallazo.jpg' (41 KB)

Preview of image 'pantallazo.jpg'

On 2012-10-15 07:49, Jaime Vives Piqueres wrote:
>    For the moment there are no compound nor constrains, just rigid meshes
> linked together because of the geometry and careful placement: the
> simulation does it's job, and the links work, even if sometimes break
> due to accuracy problems (still haven't decided if it's a bug or a
> feature).

Speaking of constraints, there's this thing I've heard about called 
Verlet integration, which I don't know how it compares directly to the 
cloth and hair simulation in Poser, but suspect it's different and 
better/faster...ever played with that at all?

--
Tim Cook
http://empyrean.sjcook.com

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> Hi all:
>
>    This is getting really dangerous... I managed to fix the segfaults on
> Koppi's bullet physics playground, and now the fun has no end. I even
> managed to add bare mesh support, as the lib3ds code was there and
> almost finished, and this opens a whole universe of "now possible"
> scenes. I don't know how I'm going to get any serious work done from now
> on...

Sounds like you're making great strides, and what you're accomplishing now is
serious enough, IMO. You're enhancing an existing tool for the creation of more
complex scenes. Will there ever be a Windows build? :)

>    For the moment there are no compound nor constrains, just rigid meshes
> linked together because of the geometry and careful placement: the
> simulation does it's job, and the links work, even if sometimes break
> due to accuracy problems (still haven't decided if it's a bug or a feature).

I was having that problem with Blender's Bullet implementation. I was trying to
use mesh links as well, but they would break sometimes. Do you have an option of
increasing the sim's accuracy? It might help.

Nice test scenes, BTW. The chain material looks a bit like lead, though (despite
the interior rust). If radiosity took into account a finish's brilliance, it
would look much better. Sometimes the only cure is to drive the diffuse down,
raise the reflection, add a tiny bumps normal, and render the scene with focal
blur.

The pearls look good as well, almost photographic. I wonder how several
concentric, translucent and refractive spheres with that pearl texture would
turn out...

Post a reply to this message

On 15-10-2012 19:38, Jaime Vives Piqueres wrote:
> On 15/10/12 19:36, Jaime Vives Piqueres wrote:
>> I don't recommend it, as you can reach insane levels of fun, and
>> mean REALLY insane.
>
>    See the attached screen-shot from the GL preview: do you see what I
> mean? ;)

This is the kind of (in)sanity that I am willing to plunge into without 
a thought...

Thomas

Post a reply to this message

On 16/10/12 01:36, Tim Cook wrote:
> Speaking of constraints, there's this thing I've heard about called
> Verlet integration, which I don't know how it compares directly to
> the cloth and hair simulation in Poser, but suspect it's different
> and better/faster...ever played with that at all?
>

   No, sorry... this is my first approach to physics simulations, and all
I know for the moment is about rigid bodies. But I've seen some awesome
cloth physics on the bullet demos, indeed... will get there someday, I
guess.

--
Jaime

Post a reply to this message

On 16/10/12 05:38, Samuel Benge wrote:
> [...] Will there ever be a Windows build? :)

   I think Koppi has successfully built it on Windows, or at least his 
building-howto seems to suggest it. But I'm guessing he is not planning 
to distribute and support binaries... will ask him anyhow, to see if we 
can at least have some unofficial binaries to distribute here.

> I was having that problem with Blender's Bullet implementation. I was trying to
> use mesh links as well, but they would break sometimes. Do you have an option of
> increasing the sim's accuracy? It might help.

   Well, I was referring to accuracy of the meshes, not the simulation: 
I discovered that just subdividing the problematic parts of the mesh was 
enough to make it more resistant to breaking. The breaking seemed to 
occur at places were both bodies didn't have vertex to collide with.

   I also discovered that size really matters too... don't make your 
objects too small. At first, I started with the default Wings3D scale of 
1-2 units, but later scaled it x10 and things seem to work better.

   One last note: I also noticed that centering the mesh model seems 
important for the mass-center of the object later on the simulation.

> Nice test scenes, BTW.

   Actually, I've rendered a lot more, mostly "piles of things" using 
existing models of mine, but didn't want to flood the newsgroups with 
them (I should update my web site ASAP, if I can get out of the playground).

> The chain material looks a bit like lead, though

   Yeah... reflections were taking too much time, so I used just 
metallic specular albedo.

> (despite the interior rust).

   That was a nice trick with a scaled spherical pattern, profiting from 
the object specific shape... no uv_mapping at all.

> raise the reflection, add a tiny bumps normal, and render the scene with focal
> blur.

   Yes, and wait until next week for the image, I guess... ;)

> The pearls look good as well, almost photographic.

#declare n_pearl=
normal{
   radial sine_wave
   frequency 36*4
   rotate 90*z
}
#declare t_pearl=
material{
   texture{
     pigment {
       aoi
       color_map{
         [0 Wheat*.5+Coral*.5]
         [1 White]
       }
     }
     normal{n_pearl .05}
     finish{
       specular albedo .25 metallic roughness .025 diffuse .75 brilliance 2
       reflection{0,1 metallic}
       conserve_energy
       irid {
           0.35
           thickness .1
           turbulence .5
         }
     }
   }
}

   Got it just by trial&error...

> I wonder how several concentric, translucent and refractive spheres with that pearl
texture would
> turn out...

   Hmmm... no time to try it now, but it's a nice suggestion that I will 
surely try out.


--
Jaime

Post a reply to this message

On 16-10-2012 12:48, Jaime Vives Piqueres wrote:
>    Well, I was referring to accuracy of the meshes, not the simulation:
> I discovered that just subdividing the problematic parts of the mesh was
> enough to make it more resistant to breaking. The breaking seemed to
> occur at places were both bodies didn't have vertex to collide with.

Good point. This is also very important with cloth simulation in Poser.

>
>    I also discovered that size really matters too... don't make your
> objects too small. At first, I started with the default Wings3D scale of
> 1-2 units, but later scaled it x10 and things seem to work better.

idem.

Thomas

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:

>    Anyhow, if you want to try it, you may need to put some hard work to
> get it compiled. The author has not wrote anything into the wiki howto,
> but seems he plans to... the problems is that he is unreachable until
> mid-december. I really don't know for sure, but I think all the
> libraries used in the project must be available for OS X, and gcc too,
> so it should be relatively easy if you already have experience compiling
> from sources on OS X.
>
>    In any case, if you manage to compile it, then contact me for the
> fixes I did to the current code on github, else you will not be able to
> do much with it.

I made a lot of progress on an OS X build last night. There are still some
missing header files, I'll keep at it. Once I get it working, I would like to
merge your changes in with the changes I needed to make for OS X, so far just
the location of header files.

Henry

Post a reply to this message

On 16/10/12 20:09, HenryW wrote:
> I made a lot of progress on an OS X build last night. There are still some
> missing header files, I'll keep at it. Once I get it working, I would like to

   Great news... seems that soon I will not be the only user of this 
app... ;)

> merge your changes in with the changes I needed to make for OS X, so far just
> the location of header files.
>

   No problem, the changes are really few, mostly on viewer.cpp. In 
fact, most of the changes were commented-out code which I removed, 
because it was not specific for the program, but for some experiments 
made by Koppi.

--
Jaime

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> On 16/10/12 20:09, HenryW wrote:
> > I made a lot of progress on an OS X build last night. There are still some
> > missing header files, I'll keep at it. Once I get it working, I would like to
>
>    Great news... seems that soon I will not be the only user of this
> app... ;)
>
> > merge your changes in with the changes I needed to make for OS X, so far just
> > the location of header files.
> >
>
>    No problem, the changes are really few, mostly on viewer.cpp. In
> fact, most of the changes were commented-out code which I removed,
> because it was not specific for the program, but for some experiments
> made by Koppi.

Another evening of keeping at it, made progress, but not there yet. I'm running
into linking errors with MIDI libraries. May be due to OS X handling sound
differently. Does the app have sounds? I'm going to comment out that code.

Henry

Post a reply to this message

On 17/10/12 15:40, HenryW wrote:
> Another evening of keeping at it, made progress, but not there yet.
> I'm running into linking errors with MIDI libraries. May be due to
> OS X handling sound differently. Does the app have sounds? I'm going
> to comment out that code.
>

   No, I think MIDI and "robots" support where added by Koppi to create
some "hard-wired" animations... I was tempted too to take it out,
but was afraid to mess up the project files. If you are successful at
it, please let me know so I can do the same.

--
Jaime

Post a reply to this message

Le 17/10/2012 15:40, HenryW a écrit :
> The first one consists of a lathe object shaped into a bulb, referenced with a
> negative scale component to create a hourglass shape then a scaled down version
> (0.995) is differenced from it. You can see banding which looks to me as if the
> glass does not have a uniform thickness.



Hint: if you scale, the object get a dilatation.

E.g. Value of 1 scaled 0.995 becomes 0.995
Value of 10 scaled 0.995 becomes 9.95
Value of 100 scaled 0.995 becomes 99.5

Width at 100 / scaled : 0.5
width at 10 / scaled : 0.05
width at 1 / scaled : 0.005

If you want a constant "horizontal" thickness, you need to translate the
curve of the lathe BEFORE rotating it. (assuming a vertical axis for the
lathe / sor )
(it's only horizontal : the slant of the curve might very well impact
the overall perception of thickness, you would need to draw a second
curve using fixed-width-discs along the first curve to really have a
uniform thickness: I can do it on a paper, but it's a nightmare to
program that for all possible curve )

Post a reply to this message

On 17-10-2012 16:02, Le_Forgeron wrote:
> Le 17/10/2012 15:40, HenryW a écrit :
>> The first one consists of a lathe object shaped into a bulb, referenced with a
>> negative scale component to create a hourglass shape then a scaled down version
>> (0.995) is differenced from it. You can see banding which looks to me as if the
>> glass does not have a uniform thickness.
>
>
>
> Hint: if you scale, the object get a dilatation.
>
> E.g. Value of 1 scaled 0.995 becomes 0.995
> Value of 10 scaled 0.995 becomes 9.95
> Value of 100 scaled 0.995 becomes 99.5
>
> Width at 100 / scaled : 0.5
> width at 10 / scaled : 0.05
> width at 1 / scaled : 0.005
>
> If you want a constant "horizontal" thickness, you need to translate the
> curve of the lathe BEFORE rotating it. (assuming a vertical axis for the
> lathe / sor )
> (it's only horizontal : the slant of the curve might very well impact
> the overall perception of thickness, you would need to draw a second
> curve using fixed-width-discs along the first curve to really have a
> uniform thickness: I can do it on a paper, but it's a nightmare to
> program that for all possible curve )
>

Well... this is an answer to Stephen's lathe problem ;-)

Thomas

Post a reply to this message

On 17/10/2012 3:02 PM, Le_Forgeron wrote:
> Le 17/10/2012 15:40, HenryW a écrit :
>> The first one consists of a lathe object shaped into a bulb, referenced with a
>> negative scale component to create a hourglass shape then a scaled down version
>> (0.995) is differenced from it. You can see banding which looks to me as if the
>> glass does not have a uniform thickness.
>
>
>
> Hint: if you scale, the object get a dilatation.
>
> E.g. Value of 1 scaled 0.995 becomes 0.995
> Value of 10 scaled 0.995 becomes 9.95
> Value of 100 scaled 0.995 becomes 99.5
>
> Width at 100 / scaled : 0.5
> width at 10 / scaled : 0.05
> width at 1 / scaled : 0.005
>
> If you want a constant "horizontal" thickness, you need to translate the
> curve of the lathe BEFORE rotating it. (assuming a vertical axis for the
> lathe / sor )
> (it's only horizontal : the slant of the curve might very well impact
> the overall perception of thickness, you would need to draw a second
> curve using fixed-width-discs along the first curve to really have a
> uniform thickness: I can do it on a paper, but it's a nightmare to
> program that for all possible curve )
>

As Thomas pointed out, you have hijacked HenryW's thread. I blame 
Thunderbird. ;-)

Yes I agree with you. I had just hoped that it wouldn't have been so 
obvious. Possibly the inner curve extends beyond the outer one at some 
points. I have moved on to using two curves.

Thank you.

-- 
Regards
     Stephen

Post a reply to this message

Stephen <mca### [at] aolcom> wrote:
> As Thomas pointed out, you have hijacked HenryW's thread. I blame
> Thunderbird. ;-)

It started as Jaime's thread for the cool work he is doing with complex objects
in Bullet Physics Playground. I hijacked it for how to compile the code on my
Mac, so I can play with Bullet Physics too :)

Henry

Post a reply to this message

On 18/10/2012 12:53 AM, HenryW wrote:
> Stephen <mca### [at] aolcom> wrote:
>> As Thomas pointed out, you have hijacked HenryW's thread. I blame
>> Thunderbird. ;-)
>
> It started as Jaime's thread for the cool work he is doing with complex objects
> in Bullet Physics Playground. I hijacked it for how to compile the code on my
> Mac, so I can play with Bullet Physics too :)
>

:-D

-- 
Regards
     Stephen

Post a reply to this message

Ok, had to finish it... tough I will revisit it for the rust work someday.

--
Jaime

Post a reply to this message

Attachments:
Download 'pov-keyring-05.jpg' (178 KB)

Preview of image 'pov-keyring-05.jpg'

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> Ok, had to finish it... tough I will revisit it for the rust work someday.

Nice work Jaime! The textures and lighting are very realistic.

Henry

Post a reply to this message

On 18-10-2012 23:27, Jaime Vives Piqueres wrote:
> Ok, had to finish it... tough I will revisit it for the rust work someday.

This is really becoming something to be made part of POV-Ray, one way or 
another. Identical to Moray for instance.

Thomas

Post a reply to this message

On 19/10/12 09:19, Thomas de Groot wrote:
> This is really becoming something to be made part of POV-Ray, one
> way or another. Identical to Moray for instance.

   You mean the Bullet Physics Library? It would be very difficult to
integrate it "natively" on the POV-Ray SDL, I guess...

   But an external tool like Koppi's Playground is not a bad idea, if it
can be made simple and practical enough. The main thing here is that you
have to "arrange" the scene in LUA, run the simulation, export to
POV-Ray, then customize the export and finally render the scene.

   The last part can (and must) be simplified enough so that you just
export, then render the scene with the appropriate #include. This can be
accomplished by just having an "attached SDL" property on each collision
object, so it exports what you want after adding the corresponding matrix.

   For the first step, describing the scene in LUA, I'm thinking that
perhaps it can be done with some clever SDL macros that will write the
corresponding LUA script for the simulation... hmmmm, just a loose idea
for the moment.

--
Jaime

Post a reply to this message

On 19-10-2012 13:30, Jaime Vives Piqueres wrote:
>    But an external tool like Koppi's Playground is not a bad idea, if it
> can be made simple and practical enough. The main thing here is that you
> have to "arrange" the scene in LUA, run the simulation, export to
> POV-Ray, then customize the export and finally render the scene.
>

Yes, that is what I was thinking about; in the same manner as Moray is 
stand alone (including views) but exports easily to POV-Ray.

Thomas

Post a reply to this message

Jaime Vives Piqueres <jai### [at] ignoranciaorg> wrote:
> Ok, had to finish it... tough I will revisit it for the rust work someday.
>

Nice textures, good placement... I like it :)

Post a reply to this message

Jaime Vives Piqueres wrote:

>   You mean the Bullet Physics Library? It would be very difficult to
> integrate it "natively" on the POV-Ray SDL, I guess...

well, it could probably be done similar to megapovs mechsim, allowing
to tag supported primitives with some metadata that adds them to the
bullet sim, and then adapting the object positions for each simulation
step / frame when rendering as an animation (or a fake animation mode
where only the final state is actually rendered).

Post a reply to this message

On 19/10/12 22:18, Samuel Benge wrote:
> Nice textures,

   That's the new albedo specular... and me using a non-default
brilliance for first time in my life. :(

> good placement...

   That's not my fault at all... ;)

   Now seriously, this keyring with keys behaves absolutely realistic on
the playground when you let it fall: sometimes it even gets one of the
keys mired into the ring! (I hate when this happens in RL and I need to
get into my house *urgently*).

> I like it :)

   Thanks!

--
Jaime

Post a reply to this message

Am 17.10.2012 16:01, schrieb Jaime Vives Piqueres:
> On 17/10/12 15:40, HenryW wrote:
>> Another evening of keeping at it, made progress, but not there yet.
>> I'm running into linking errors with MIDI libraries. May be due to
>> OS X handling sound differently. Does the app have sounds? I'm going
>> to comment out that code.
>>
>
>    No, I think MIDI and "robots" support where added by Koppi to create
> some "hard-wired" animations... I was tempted too to take it out,
> but was afraid to mess up the project files. If you are successful at
> it, please let me know so I can do the same.

MIDI-driven animations, maybe? Sounds like too much fun to strip it...

Post a reply to this message

On 26/10/12 23:25, clipka wrote:
> MIDI-driven animations, maybe? Sounds like too much fun to strip
> it...

   I think it is for use on the opposite direction... to launch sounds
when objects collide. Which could be fun too...

--
Jaime

Post a reply to this message